A Discipline for Software Engineering

Transcription

1 A Discipline for Software Engineering (Humphrey, (Humphrey, 1995) 1995) Introduction AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 1 Outline Software Development: Craft or Discipline? How SE is taught Humphrey s book s approach AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 2 1

2 Components of a Disciplined SE Process (Humphrey, (Humphrey, 1995, 1995, p. p. ix; ix; Clark, Clark, 1996) 1996) Defect Management Comprehensive methods for: Planning Tracking Analysis Integrated control mechanisms completes the project management tool set Humphrey does not refer to this AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 3 Software Engineering - Craft or Engineering Discipline? Examples Automated luggage system delays new airport opening by 1 1/2 years at $1.1 million per day in interest and operating costs. (Gibbs, 1994, p. 86) California DMV pulls the plug after a 7- year, $44.3-million debacle. Over 6 times the original budget. Over 24% of projects over 5000 function points in size are canceled. (Jones, 1995, p. 3, and Gibbs, 1994, p. 88) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 4 2

3 Software Engineering - Craft or Engineering Discipline? It s like musket making was before Eli Whitney. (Brad Cox, quoted in Gibbs, 1994, p. 87) Despite 50 years of progress, the software industry remains years - perhaps decades - short of the mature engineering discipline needed to meet the demands of an information-age society. (Gibbs, 1994, p. 87) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 5 Software Engineering - Craft or Engineering Discipline? Why do we have these problems? Is Software Engineering REALLY an engineering discipline, or is it still just a craft? AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 6 3

4 Prospects for an Engineering Discipline of Software (Shaw, 1990) Origins of the Term Software Engineering 1968 NATO workshop Def: Engineering (p. 15) a set of current best practices for development. the disciplined application of scientific knowledge to resolving conflicting constraints and requirements for problems of immediate, practical significance. AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 7 Prospects for an Engineering Discipline of Software (Shaw, 1990) Evolution of an Engineering Discipline AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 8 4

5 Prospects for an Engineering Discipline of Software (Shaw, 1990) Evolution of Civil Engineering AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 9 Prospects for an Engineering Discipline of Software (Shaw, 1990) Evolution of Chemical Engineering AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 10 5

6 Prospects for an Engineering Discipline of Software (Shaw, 1990) Evolution of Software Engineering AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 11 Measuring and Managing Information is much easier to store, describe, and manipulate than knowledge. (p. 61) The level of knowledge that a process has reached determines how a process should be controlled, whether and how it can be automated, the key tasks of the workforce, and other major aspects of its management. (p. 61) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 12 6

7 Measuring and Managing Examples Chaparral Steel doubled mill output used original electric furnace and caster Semiconductor manufacturers routinely obtain 80% yield after only a few years, when starting with 40% minimal capital investment changed mfg. process (procedures, control adjustments, raw materials recipes, etc.) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 13 Measuring and Managing Definition: Technological Knowledge Technological knowledge [is] understanding the effects of the input variables on the output. (p. 62) Diagram of a Process AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 14 7

8 Measuring and Managing Eight Stages of Knowledge (p. 63, 64) Name Comment Typical Knowledge Form 1 Complete Ignorance Nowhere 2 Awareness Pure Art Tacit 3 Measure Pretechnological Written 4 Control of the mean Scientific method feasible Written & embodied in hardware 5 Process capability Local recipe Hardware and operating manual 6 Process characterization Tradeoffs to reduce costs Empirical equations (numerical) 7 Know why Science Scientific formulas & algorithms 8 Complete knowledge Nirvana AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 15 Measuring and Managing Evolution of Knowledge and Performance (p. 65) Knowledge about [a] process and how to run it is incomplete and develops gradually through various kinds of learning. (p. 61) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 16 8

9 Measuring and Managing Knowledge-based Learning and Organizational Improvement AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 17 Measuring and Managing Balance Between Degree of Procedure & Stage of Knowledge Automation of a large, complex, poorly understood, conventional manufacturing process leads to a large, complex, poorly understood, unreliable, expensive, and automated manufacturing process. (p. 67) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 18 9

10 Measuring and Managing Applying Stages of Knowledge 1.Understand how much you know and don t know What are the important variables? At what stage are they? Which ones give most leverage at the next stage? How can you manage the process well at these stages? Are your management methods consistent with knowledge levels? How should you handle the inevitable variables which you know less about but which are still important? AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 19 Measuring and Managing Applying Stages of Knowledge (cont.) 2.Understand and manage the locations of knowledge 3. Be wary of de-skilling the workforce and freezing the process 4. Learn carefully and systematically AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 20 10

11 Measuring and Managing Lord Kelvin, in the 1890s: When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind: it may be the beginning of knowledge, but you have scarcely, in your thoughts, advanced to the stage of science. (quoted on p. 72) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 21 Prospects for an Engineering Discipline of Software (Shaw, 1990) 5 Steps to Become True Engineering (p ) 1.Understand the nature of expertise. 2.Recognize different ways to get information. 3.Encourage routine practice. 4.Expect professional specializations. 5.Improve coupling between science and commercial practice. AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 22 11

12 Legal Status of Software Engineering (Jones, 1995) Software engineering is not one of the 36 engineering professions recognized and licensed in the United States states have laws... that prohibit anyone who is not licensed from using the term engineer in describing his occupation and work. (Jones, 1995) Tennessee now actively prohibits the use of software engineering in business literature and advertising. (Jones, 1995) The state of Texas has forced universities to stop offering master s degrees in software engineering. (Jones, 1995) ABET has asked the IEEE to look at software engineering accreditation. ABET, the Accreditation Board for Engineering and Technology, is the legally-recognized accrediting agency for all engineering, engineering technology, and engineering-related education in the U.S.A. (Gillespie, 1997) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 23 Legal Status of Software Engineering (Jones, 1995) What makes an engineering profession? Well-defined body of knowledge, often many sub-specialties Academic curricula to transfer the knowledge and prepare students to pass qualifying exams Qualifying exams certifying minimal competence for practicing the profession Continuing education, to maintain currency Code of ethics Strong professional associations Recognized canon of standard practices - malpractice may be evaluated against these Methods for monitoring and addressing malpractice Liability insurance AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 24 12

13 Software Engineering - Craft or Engineering Discipline? So what is software engineering? The establishment and use of sound engineering principles in order to obtain economically software that is reliable and works efficiently on real machines. (Fritz Bauer, quoted in Pressman, 1992, p. 23.) The development, on schedule and within budget, of high quality software that meets the users and organization s needs. (Turk) AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 25 Software Engineering Success (Turk, 1995) Factors Affecting Success: Technological Personal Group Managerial Geographical Cultural Measures of Success: Project vs. Product Financial vs. Acceptance / Use AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 26 13

14 So, How is SE Taught? Typically, learn programming languages on toy problems. vs. Learn industrial practices on toy problems. Initial learning will require scaled-down ( toy ) problems. However, one can still learn industrial practices on these scaled-down problems. This is much better than learning scaled-down practices which don t transfer to large development projects. AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 27 This Book s Approach (Humphrey, (Humphrey, 1995, 1995, p. p. x) x) Learn how to: Make accurate plans Estimate the accuracy of these plans Track actual performance against these plans These techniques are the basis for implementing controls which allow software projects to be well managed. AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 28 14

15 Exercises To Teach These Skills Approximately 10 exercises Approximately 5-10 hours for each exercise. Each exercise teaches new skills and further expands your process capabilities. AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 29 Keys to Success in Implementing the PSP (Humphrey, (Humphrey, 1995, 1995, p. p. xiii, xiii, xiv) xiv) Most industrial organizations who successfully implement the PSP training program are at CMM level 2 or greater. (Although I think that introducing the PSP can be a tool for getting organizations TO levels 1 and 2 ) Visible senior management support. (Project managers will support it if senior manager does.) Voluntary engineer participation. Apply the techniques to text s problems, not to work at first. (Turn-around and feedback quicker.) Work problems as you read the chapters. AU INSY 560, Singapore 1997, Dan Turk Humphrey Preface - slide 30 15